Transmission of Financial Crises and Contagion:

A Latent Factor Approach

Mardi Dungey, Renee A. Fry, Brenda Gonzalez-Hermosillo, and Vance L. Martin

CERF Monographs on Finance and the Economy

Provides a thorough review of the methods developed by the authors for measuring the effects of contagion during financial crises. No other existing work measures the extent of contagion effects beyond simply detecting its existence. Code for implementation is available.

Transmission of Financial Crises and Contagion:

A Latent Factor Approach

Mardi Dungey, Renee A. Fry, Brenda Gonzalez-Hermosillo, and Vance L. Martin

CERF Monographs on Finance and the Economy

Description

Financial crises often transmit across geographical borders and different asset classes. Modeling these interactions is empirically challenging, and many of the proposed methods give different results when applied to the same data sets. In this book the authors set out their work on a general framework for modeling the transmission of financial crises using latent factor models. They show how their framework encompasses a number of other empirical contagion models and why the results between the models differ. The book builds a framework which begins from considering contagion in the bond markets during 1997-1998 across a number of countries, and culminates in a model which encompasses multiple assets across multiple countries through over a decade of crisis events from East Asia in 1997-1998 to the sub prime crisis during 2008. Program code to support implementation of similar models is available.

Transmission of Financial Crises and Contagion:

A Latent Factor Approach

Mardi Dungey, Renee A. Fry, Brenda Gonzalez-Hermosillo, and Vance L. Martin

CERF Monographs on Finance and the Economy

Preface vii
Acknowledgements ix
1 Introduction 1
2 Review of the Empirical Literature 7
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Defining Contagion . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 AModel of Interdependence . . . . . . . . . . . . . . . . . . . 10
2.4 An EmpiricalModel of Contagion . . . . . . . . . . . . . . . . 12
2.4.1 Bivariate Testing . . . . . . . . . . . . . . . . . . . . . 13
2.4.2 Multivariate Testing . . . . . . . . . . . . . . . . . . . 15
2.4.3 Structural Breaks . . . . . . . . . . . . . . . . . . . . . 16
2.4.4 Using Just CrisisData . . . . . . . . . . . . . . . . . . 17
2.4.5 Autoregressive and HeteroskedasticDynamics . . . . . 17
2.5 Correlation and Covariance Analysis . . . . . . . . . . . . . . 19
2.5.1 Bivariate Testing . . . . . . . . . . . . . . . . . . . . . 20
2.5.2 Alternative Formulation . . . . . . . . . . . . . . . . . 22
2.5.3 Multivariate Testing . . . . . . . . . . . . . . . . . . . 24
2.5.4 Endogeneity Issues . . . . . . . . . . . . . . . . . . . . 27
2.5.5 Relationship withOtherModels . . . . . . . . . . . . . 29
2.6 Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.6.1 Stylized Facts . . . . . . . . . . . . . . . . . . . . . . . 34
2.6.2 Implementation Issues . . . . . . . . . . . . . . . . . . 34
2.6.3 Contagion Testing . . . . . . . . . . . . . . . . . . . . 36
2.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3 Contagion in Global Bond Markets 41
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.2 Background of Events and Propositions . . . . . . . . . . . . . 43
3.2.1 Stylized Facts . . . . . . . . . . . . . . . . . . . . . . . 43
3.2.2 Propositions . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.4 AFactorModel of Bond Spreads . . . . . . . . . . . . . . . . 48
3.5 EstimationMethod . . . . . . . . . . . . . . . . . . . . . . . . 54
3.6 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.A Data Definitions and Sources . . . . . . . . . . . . . . . . . . 62
3.B Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . 62
3.C Unit Root Tests . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.D EstimatedGARCHModels . . . . . . . . . . . . . . . . . . . . 63
4 Contagion in Global Equity Markets 65
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.2 AModel of Financial Turmoil . . . . . . . . . . . . . . . . . . 66
4.2.1 ABenchmarkModel . . . . . . . . . . . . . . . . . . . 67
4.2.2 AModel Incorporating Contagion . . . . . . . . . . . . 69
4.3 Empirical Issues . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.3.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.3.2 GMMEstimator . . . . . . . . . . . . . . . . . . . . . 74
4.4 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.4.1 Parameter Estimates . . . . . . . . . . . . . . . . . . . 75
4.4.2 Volatility Decompositions . . . . . . . . . . . . . . . . 77
4.4.3 Structural Break Tests . . . . . . . . . . . . . . . . . . 78
4.4.4 Robustness Checks . . . . . . . . . . . . . . . . . . . . 78
4.4.5 Comparison with BondMarket Transmissions . . . . . 79
4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5 Are Crises Alike? 83
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.2 AModel of Contagion . . . . . . . . . . . . . . . . . . . . . . 85
5.3 Empirical Factor Specification . . . . . . . . . . . . . . . . . . 89
5.3.1 Noncrisis Specification . . . . . . . . . . . . . . . . . . 90
5.3.2 Crisis Specification . . . . . . . . . . . . . . . . . . . . 91
5.4 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.4.1 Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.4.2 CrisisDates . . . . . . . . . . . . . . . . . . . . . . . . 97
5.5 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . 99
5.5.1 Evidence of Contagion . . . . . . . . . . . . . . . . . . 100
5.5.2 Comparison of Contagion Channels Across Crises . . . 103
5.5.3 Testing the Channels of Contagion . . . . . . . . . . . 104
5.6 Robustness Checks andAdditional Tests . . . . . . . . . . . . 105
5.6.1 CrisisDating SensitivityAnalysis . . . . . . . . . . . . 105
5.6.2 ConditionalMoment Tests . . . . . . . . . . . . . . . . 107
5.6.3 Structural Break Tests . . . . . . . . . . . . . . . . . . 107
5.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
5.A Model Derivations . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.A.1 Optimal PortfolioWeights . . . . . . . . . . . . . . . . 111
5.A.2 Informed Conditional Expectations . . . . . . . . . . . 112
5.A.3 Uninformed Conditional Expectations . . . . . . . . . . 113
5.A.4 Excess Returns Equation . . . . . . . . . . . . . . . . . 115
5.B Data Sources and Definitions . . . . . . . . . . . . . . . . . . 118
5.C Additional VarianceDecompositions . . . . . . . . . . . . . . 118
6 Characterizing Global Risk 119
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
6.2 AModel of Risk Premia . . . . . . . . . . . . . . . . . . . . . 121
6.2.1 Model Specification . . . . . . . . . . . . . . . . . . . . 122
6.2.2 Identifying RiskQuantities . . . . . . . . . . . . . . . . 124
6.2.3 Estimation . . . . . . . . . . . . . . . . . . . . . . . . . 126
6.3 Empirical Estimates . . . . . . . . . . . . . . . . . . . . . . . 126
6.3.1 RiskQuantities . . . . . . . . . . . . . . . . . . . . . . 126
6.3.2 Risk Prices . . . . . . . . . . . . . . . . . . . . . . . . 128
6.4 Historical Decomposition of Risk Premia . . . . . . . . . . . . 129
6.4.1 Benchmark Spread Estimates . . . . . . . . . . . . . . 130
6.4.2 Global Risk Factor Estimates . . . . . . . . . . . . . . 130
6.4.3 Country Risk Factor Estimates . . . . . . . . . . . . . 131
6.4.4 ContagionRisk Factor Estimates . . . . . . . . . . . . 132
6.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
6.A Data Definitions and Sources . . . . . . . . . . . . . . . . . . 134
6.B CrisisDates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
7 Conclusions 135
Bibliography 141

Transmission of Financial Crises and Contagion:

A Latent Factor Approach

Mardi Dungey, Renee A. Fry, Brenda Gonzalez-Hermosillo, and Vance L. Martin

CERF Monographs on Finance and the Economy

Author Information

MD: Professor of Economics and Finance, University of Tasmania
RF: Research Associate, Centre for Financial Analysis and Policy
BGH: Deputy Division Chief of Global Financial Stability, Monetary and Capital Markets Department, International Monetary Fund
VM: Professor of Econometrics, University of Melbourne

Transmission of Financial Crises and Contagion:

A Latent Factor Approach

Mardi Dungey, Renee A. Fry, Brenda Gonzalez-Hermosillo, and Vance L. Martin

CERF Monographs on Finance and the Economy

Reviews and Awards

"Examining the role of contagion during financial crisis is an important and difficult task. This book provides a coherent framework for identification, estimation and testing for the presence of contagion. The importance of the framework is illustrated by a number of applications to many recent finiancial crises. I highly recommend it for students, practitioners and scholars interested in linkages between asset markets."--Carlo A. Favero, Professor of Financial Econometrics, Bocconi University

"Policymakers, observers and academics are still struggling to grasp how developments in one particular country (the United States) and one specific, rather narrow market segment (the subprime mortgage market) could turn in 2007-09 into the worst global financial meltdown since the Great Depression. The book makes an important contribution to our understanding of the underlying transmission mechanism that has turned so many individual, isolated events over the past decades into systemic crises. It does so by developing a powerful and elegant conceptual framework, and by illustrating how this framework helps trace the contagious spread of crises. The book could not be more timely and stresses the need for policy cooperation at the global level, and for policymakers to better take into account the rising global interconnectedness when taking decisions."--Marcel Fratzscher, Head of International Policy Analysis Division, European Central Bank

"Despite the large and growing research on financial crises and spillovers, there was much confusion about empirical modeling and testing of links across financial markets. Not anymore. This book presents a new coherent framework for testing and quantifying contagion. At the core of this new framework is the use of latent factor techniques. I highly recommend reading chapter 2 for interpreting, in a new light, previous results in the literature as well as new extensions to deal with structural breaks and endogeneity issues. The case studies throughout the book are very helpful in illustrating the techniques. This is a great book. In the midst of the Subprime-Euro-Greek crisis, this book is a must read."--Graciela L. Kaminsky, Professor of Economics and International Affairs, George Washington University